Modern wireless devices have a large range and therefore may be in contact with many other wireless devices at any particular time. Thus, if a device is seeking to transfer data to or from another device, it may have a large number of devices from which to select the desired wireless device. Additionally, there is the possibility that another device within range may interfere with, or breach the security of, a particular wireless device. To minimize the potential for a security breach or interference issues, a formal connection process is often initiated between wireless communications devices.
Many RF communications systems employ frequency diversity to minimize interference. This helps design robust systems, but as with many technology designs, this often results in design tradeoffs, and in particular, introduces latency: if there are N channels available, and to detect if a device is present on any one channel takes a maximum time τ, then there is a connection-pairing latency up to Nτ between two devices. In a typical multiple-frequency narrowband communication system, a base station designates a frequency (channel) for communication with a particular device, the two devices perform a handshake protocol to establish communications over the designated channel, and typically other devices are required to avoid the channel(s) that are in-use by other devices, e.g., by detecting channels that are in-use or by using a designated communication path for establishing communications with the base station.
To improve latency, the time τ may be reduced such that NT is imperceptible to a human. However, reducing the time τ may be a challenge because frequency hop time typically is set by the loop filter of the phase-locked loop (PLL) of the radio chip, and this may be constrained for reasons such as PLL noise performance requirements. One may also design a media-access control (MAC) layer protocol stipulating new devices initialize a network session with an a priori channel selection. This may work in a peer-to-peer environment if there is sufficient signal to interference-plus-noise ratio (SINR) at this specific channel to allow communication to occur, although it may not be ideal for a base-station-to-device model, as the base station may spend valuable time on this channel while not using other channel(s) for devices already connected. At the MAC level, there is also the issue of how much time a radio should spend trying to connect to other devices versus how much time should be spent communicating with devices that are already connected. Therefore, the frequency hopping design may be less robust than other designs when attempting to maximize throughput, while minimizing latency for communications. Many wireless protocols have different design tradeoffs for the optimization of throughput and latency. For example, in wide area networks using wireless standards such as 3G and 4G, the initial association latency may be large, as device to cell tower communication may persist for a long time and base station protocols have appropriate protocols for base station hand off as the wireless device moves around in an environment. In the IEEE 802.11 Wi-Fi standard, association times can be several seconds due to the relative low duty cycle of broadcast Beacon commands (typically every 100 ms), and the number of channels (11 primary Wi-Fi channels in the US in the 2.4 GHz frequency range). In the Bluetooth Low Energy (BTLE) standard, 3 reserved channels are provided out of 40 total for communication initiation to improve latency.
Some wireless devices can transfer data via infrared communication ports or with radio frequency (RF) data transfer. Microwave technologies such as Bluetooth and Wi-Fi allow non-line-of-sight device-to-device communication. However, due to security concerns, these technologies require a set-up process in which a device must be added to the network. Although near-field-communication (NFC) can be used to exchange data between devices without adding a device to network, it functions only at a distance of 10 cm or less, and in practice this distance is 4 cm or less. Near field communication (NFC) is a radio frequency identification (RFID) protocol that operates with RF fields in the near-field, operating at 13.56 MHz. It is a superset of the ISO14443 and ISO18092 protocols, including security features such as elliptic curve cryptography (ECC) and the advanced encryption standard (AES). NFC is also used to exchange configuration information for other wireless standards such as Bluetooth and Wi-Fi; many Bluetooth headsets now include NFC tags for provisioning purposes. The gesture of placing a NFC headset near a phone now carries digital associative meaning.
There are disadvantages of relying on NFC to configure another wireless system. First, there is additional cost associated with providing NFC function in a wireless device. In the situation where a mobile phone carries the NFC radio, this can add geometric volume for circuitry and antennas that may pose tradeoffs with other components in the system, such as battery life, design and wireless functionality. Second, as NFC uses two physical layer protocols that must be time-sequenced, some transactions involving security can take appreciably longer than standard user interface latency of less than 10 ms. Finally, not all interactivity for establishing communications can be done within 0-10 cm; the range of manipulation by a person is limited to reach of the arms, which is typically 0.3-1 m. There are also cases in which a person is stationary but can see another person or object he/she potentially would like to connect to; he/she could walk to this location, but the wireless device in principle could allow almost imperceptible time to exchange information, obviating the need for ambulation. The operation of a television is an example of this situation, but the communication is generally handled by connectionless infrared protocols, or pre-associated devices based on Bluetooth or Wi-Fi.
One method of exchanging data is through passwords or secret keys. Another method of exchanging data with lower burden on the user is through time synchronization of an interaction, in which the users of two or more devices press a button to open a small security hole for a short window of time and exchange security keys. Some existing mobile devices can exchange accelerometer signals recorded when users bump their phones together. The exchange of accelerometer signals allows the devices to then exchange information.
FIG. 1 is a diagram of a prior art communications system 100 to which various embodiments of the invention may be applied. (Similarly various embodiments of the invention may be applied to the prior art arrangements illustrated in FIGS. 2A, 2B, 2C, 2D, 3A, 3B, 3C, and 4 described below.) The communication system 100 includes a base station 102 and multiple wireless devices 104a, 104b, 104c and 104d. The base station itself may be of the same device type as the wireless devices (e.g., a wireless device may act as a base station for some or all communication transactions). The base station transmits an RF signal 106 received by the wireless communication devices. According to one embodiment, the base station 102 is connected to a power source. The power source may be an electrical outlet, battery, or other electrical source. The base station 102 may also include one or more network interfaces for coupling to one or more wired or wireless networks, including, for example, a Local Area Network (LAN), a Wireless Local Area Network (WLAN), a Wide Area Network (WAN), a cellular network or a Public Switched Telephone Network (PSTN). According to various embodiments, wireless communication devices 104a-104d may include one or more mobile phones, iPhones, headphones, headsets (including a microphone and earphone), music players, iPods, personal digital assistants, iPads, tablets, laptops, computers, cameras, or other types of devices.
FIG. 2A is a diagram of a pair of prior art communication devices communicating to each other via a narrowband communication system 200. Each device is using a local oscillator (LO) in the transmitter and receiver path, specifically, one device is using LO1 205 and the other device is using LO2 210. One of the devices may be (or act as) a base station and the other may be (or act as) a wireless device. Since the receiver of each device has a finite bandwidth, the devices share a channel plan indicating, within the accuracy and precision of their local clocks that generate their local oscillators (LOs), what channel two devices will share to communicate with each other. There may be multiple communication channels, to allow bandwidth sharing and channel robustness from external interferers. For example, the 2.4 GHz Wi-Fi band from 2.403 GHz to about 2.483 GHz has somewhere between 11 and 14 overlapping channels. When two devices share the same channel, they are able to communicate with each other with high data rate corresponding to a large channel bandwidth, and depending on the protocol, low latency as well. When two devices do not share the same channel, some amount of time elapses while they switch their local oscillators to the same channel. For example, in some systems, the time may vary from about 100 μs is to several milliseconds. There also may be protocol-level latency associated with changing channels, such as the latency associated with beacon or advertising frames of data. If the transceiver of a device switches several times before occupying the same channel as the other device, this process may take several milliseconds to seconds.
FIG. 2B is an example showing a time domain waveform of an RF signal and waveforms of the same signal after detection using respectively a broadband detector 235b and a narrowband detector 240b. In this example, a transmitter from 205 modulates data using phase reversal amplitude shift keying (PR-ASK) modulation, typically used by the GS1/EPCG Global Gen2 or ISO18000-6C RFID protocol. The 0 and 1 bits are encoded using different time durations, with the bit sequence 010011 encoded in this example. The RF modulation 225b and a zoomed in version 230b show the RF cycles at 915.0 MHz. The RF signal 225b from the transmitter signal is detected both by a transceiver 210 using a broadband detector and by a separate, independent transceiver 210 using a narrowband detector. The broadband detector waveform 235b results from using a diode and single-pole low-pass filter envelope detector; the detected signal 235b from the broadband detector is similar to the source waveform 220b, but has some unfiltered and distorted parts of the original RF signal as a result of the nonlinearity of the detector and the characteristics of the single-pole filter. Nevertheless, the signal fidelity is more than adequate to extract the original bit sequence. The waveform 240b, resulting from application of the original RF signal to the narrowband detector, is an undistorted representation of the original waveform 220b, because the local oscillator (LO) of the narrowband detector closely matches the source (they are on the same channel). The high signal fidelity allows ready extraction of the original bit sequence.
FIG. 2C is an example similar to FIG. 2B, with the addition of a continuous wave (CW) interfering signal that is one channel higher (+10 MHz) than the transmitter. The amplitude of the interfering signal is ½ (that is, 3 dB below) that of the transmitter signal. The source signal with interference 225c is shown relative to the original envelope 220c from the transmitter. The broadband detected signal 235c shows additional distortion, as the entire waveform including interference passes through the nonlinearity and single-pole filter. If the detector has a variable gain and/or variable threshold, the original bit sequence might still be extracted, but with potentially higher bit error probability. As the interferer amplitude increases above ½ or −3 dB, the receiver will no longer be able to extract the bit pattern. But, as interferers remain further away from the broadband RF receiver than the transmitter, it is possible for data to be reliably extracted from the broadband RF receiver, as is done with RFID tags based on the GS1/EPCG Global Gen2 or ISO18000-6C RFID protocol. With a channel filter on the narrowband detector, the narrowband detected signal 240c is identical to 240b, and the interference is eliminated from the narrowband detector.
FIG. 2D is an example similar to FIG. 2B, except the LO of the transmitter from 205 is changed to the next channel, 10 MHz above 915.0 MHz. For the broadband detector, the detected signal is relatively independent of the LO frequency, and therefore the detected envelope 235d is essentially equivalent to 235b. The signal fidelity of the waveform 235d is more than adequate to extract the original bit sequence. For the narrowband detector with an LO of 915.0 MHz, as in the original example, the channel filter rejects the transmitted signal due to a LO mismatch. The waveform 240d is absent. It is possible there may be sufficient dynamic range in the narrowband receiver to recover the original data sequence, but major structural changes in the modem of the receiver may be required to manage the significant LO mismatch. There is an explicit tradeoff between channel rejection and the instantaneous acquisition of information between narrowband devices.
FIG. 3A is a time sequence diagram of a prior art arrangement wherein a client is connecting to an access point (AP) using the 802.11 protocol, using either Direct Sequence Spread Spectrum (DSSS) or Orthogonal Frequency Division Multiplexed (OFDM) modulation. The base station and client both operate narrowband transceivers that utilize oscillators and must choose a channel in the 2.4 and/or 5 GHz band to operate in. The base station sets its local oscillator to a CH1, which will be a specific frequency that is specific to a specific country of operation. For example, in the United States, there are 11 channels, starting at 2412 MHz (CH1) up to 2462 MHz (CH11). In the example shown in FIG. 3A, the client, trying to connect to an AP, sets its local oscillator LO2 to CH5. Since the Beacon packet from the AP, which are used to provide information to clients to connect to the AP, are on CH1, the client at CH5 does not see the Beacon packet. If there are no collisions on the channel, the AP will typically transmit Beacon packets every 100 ms. If there are collisions, this interval could be a multiple of 100 ms. The client, not seeing a Beacon frame, must jump to another channel to find a Beacon frame. In the worst possible case without collisions, with 11 channels, going to the same channel of the AP could take up to 11.times.100 ms=1.1 s, with an average of 6.times.100 ms=600 ms. When collisions are considered, this is the reason it can take several seconds for a client to see an intended AP. In a client to client (or peer to peer) model such as Wi-Fi Direct, the same structure of establishing data communications is required, as one client must play the role of AP and the other client must match to the channel of the other client. If multiple users greater than 2 would like to connect, the time for all clients to be connected can grow substantially.
FIG. 3B is a time sequence diagram of a prior art arrangement wherein a peripheral is connecting to a central system using the BTLE protocol, where the protocol specifies FHSS for the channel sharing algorithm. In this protocol, three of the total 40 channels in the range of 2402-2480 MHz are advertising channels for other devices, while the remaining 37 channels are for data. In the example shown, the central system sets its local oscillator to channel 38 or 2426 MHz, one of the advertising channels, while the peripheral tries to establish communications on channel 37 or 2402 MHz. The peripheral provides an ADV_DIRECT_IND packet to look for a central system to establish communications with, but because the two narrowband devices are not on the same channel, they are not able to see each other. By being on the incorrect channel, up to 10 ms could elapse before the peripheral switches channels. Then the peripheral chooses another channel, either in a static algorithmic or table-driven way, to channel 38. This channel is now the same channel as the central system, and therefore the ADV_DIRECT_IND message sent by the peripheral can be heard by the central system if there is a sufficient SINR. The central system responds with a SCAN_REQ response, and is now able to send packets to the peripheral. In this example, the time for the peripheral to connect to the central system is under 20 ms, but in general, with 3 advertising channels, connection time could be under 10 ms, under 20 ms, or under 30 ms. With a connection time on average of 20 ms, this is a short time on human perceptible scales, but could potentially be shorter to allow more data to be transmitted during this interval.
FIG. 3C is a time sequence diagram of a prior art arrangement wherein a GS1/EPCG Global Gen2 or ISO18000-6C RFID reader is communicating with a Gen2 or ISO18000-6C RFID tag using a broadband transceiver on the tag. The tag is capable of operating over a worldwide frequency range of 860-930 MHz. The tag can either be powered by the RF field itself, termed a passive tag, or a local power source such as a battery, and it is termed a semi-passive RFID tag. The choice of power source in the embodiment of the Gen2 of ISO18000-6C protocol does not change the timing of the system, but increases the receiver sensitivity of the tag, enabling longer distance communications. Regardless of the power source, an RFID reader communicates with a RFID tag using amplitude modulation, and the tag communicates to the RFID reader using backscatter amplitude modulation. In some embodiments, the amplitude modulation is double sideband amplitude shift keying (DSB-ASK), phase reversal amplitude shift keying (PR-ASK) or single sideband amplitude shift keying (SSB-ASK). When the tag backscatters, the RFID tag generates a reflection of a partial component of the interrogating RF wave. By varying the impedance of the circuitry presented to its antenna, the RFID tag can modulate in a time-sequenced manner the amount of the partial component to communicate information. The receiver of the RFID reader is capable of extracting this partial component as an amplitude shift keyed signal. In a backscatter system, the tag does not generate or use its own local oscillator (LO); it simply communicates ASK data on the RF wave originating from the reader, as described above. This means there is no carrier synchronization required, but the tradeoff is that the path loss from the reader to the tag and back is at least double the path loss of a traditional active radio system. Therefore, the explicit tradeoff of a backscatter system with an active transmitter is that the path loss is double for the backscatter system, but the latency on any channel is significantly lower on average than an active radio system.
In the example shown in FIG. 3C, the RFID tag is able to extract power from the RF field of the reader, and therefore the Power State replaces the Local Oscillator state in this time sequence. Immediately after the reader has settled its local oscillator, it transmits CW for a time required of the protocol to turn the Power State of the tag to ON from the OFF state. The reader can then immediately modulate its RF transmitted wave to send data to the RFID tag and the tag can interpret the data. Finally, the tag processes the information sent by the reader and responds with a backscatter response to the reader information. In the EPCG/GS1 protocol, the time that a reader can communicate with a tag varies with the bit time of the transmitter and the bit time of the tag. For the example where the bit-0 time, or Tari is 6.25 μs, and the Backscatter Link Frequency (BLF) of the RFID tag is 640 KHz with FM0 modulation, the time for the reader to obtain 96-bits of information in addition to a 16-bit random number, 16-bits of protocol control bits and a 16-bit cyclic redundancy check (CRC) is approximately 2.5 ms, with each incremental packet from an RFID tag being approximately 1.0 ms. If the BLF is 400 kHz, this time is approximately 2.7 ms, with each incremental packet from an RFID tag being approximately 1.2 ms. If the BLF uses a Miller modulation with M=4 and a BLF of 256 kHz, the time is approximately 4.8 ms, with each incremental packet from an RFID tag being approximately 3.2 ms. In practice, if there are many tags in the field, the rate of incremental packet acquisition can be slowed down by the efficiency of a slotted ALOHA protocol, typically e, the base of the natural logarithm, or 2.72 times slower than the numbers described above. Also in practice, channel noise may slow down the rate of acquisition of data from the RFID tags, as would be true of any RF protocol. Nevertheless, in practice, the connection time for a broadband receiver system used in this example is capable of being significantly faster than the Wi-Fi or Bluetooth examples above.
FIG. 4 is a schematic diagram of the transceiver portion of a conventional mobile communications device 400. On the transmitter side, a modem 450 generates a set of digital signals that are converted into an analog baseband signal by the transmit baseband 425. After digital-to-analog conversion in the transmit baseband 425, low pass filtering may be implemented in this block. An I&Q modulator 420 mixes the local oscillator signal 430 with the transmit baseband signals, typically combines them into a single output, to produce modulation at the intended radio frequency. The I&Q modulator 420 may be composed of analog mixers, buffers, amplifiers, and filters. This signal is filtered, amplified 415 and then switched 410 through one or more antennas 405. On the receive side, a received signal comes through one or more switched antenna elements 410, which are amplified by a low-noise amplifier (LNA) 435, and then converted to baseband via an I&Q demodulator 440. The LNA 435 and I&Q demodulator 440 may optionally include a peak detector for channel power measurements or for automatic gain control (AGC). Like the I&Q modulator 420, the I&Q demodulator 440 may be composed of analog mixers, buffers, amplifiers, and filters. The generated I&Q analog baseband signals are processed by a series of amplifiers, analog-to-digital converters and processed using a series of digital operations that make up the receive baseband 445, to produce a digital stream that is received by the modem MAC 450. Many types of radios such as Bluetooth, Wi-Fi, GSM, RFID readers may operate in this manner at a high level, but perhaps with multiple independent transmit and receive subcomponents.